Introducing an explicit Government institution in ICES model

Transcription

1 Research Papers Issue RP0243 December 2014 CIP - Climate Impacts and Policy Division Introducing an explicit Government institution in ICES model By Elisa Delpiazzo Climate impacts and policy division, CMCC elisa.delpiazzo@cmcc.it Gabriele Standardi Fondazione Eni Enrico Mattei, FEEM and Climate impacts and policy division, CMCC gabriele.standardi@cmcc.it SUMMARY A challenging issue in global applied Computable General Equilibrium (CGE) models concerns the formalization of the government agent. Examples are the GTAP based models (as ICES), where there is not an explicit government but it is considered as part of a broader regional household. Here, a tax cut stimulates more public consumption in the economy as no public budget constraint is modelled. The aim of this paper is to present the methodology to explicitly introduce the public budget constraint and to make a clear link between tax receipts and public expenditures. Our work focuses both on the database and the model. We insert additional information to complete the GTAP database for the part related to the public finance. In the model we change the macro-economic closure between savings and investments and we link taxes and expenditures allowing the government to create debt. Keywords: CGE models, public sector, Government budget. JEL: C68, D58, H60 This report represents the Deliverable P62b developed within the framework of Work Package of the GEMINA project, funded by the Italian Ministry of Education, University and Research and the Italian Ministry of Environment, Land and Sea.

2 CMCC Research Papers SUMMARY Introduction 3 2. Modelling government in global real CGE models GTAP model GLOBE model LINKAGE, ENV-Linkage, and ENVISAGE models MyGTAP 9 3. The Regional household in the ICES model 10 4 The new Government in the ICES model An explicit government account Introducing intra- regional transfers Introducing inter- regional transfers 23 5 Building an extended database for the ICES model The starting data to build the dataset Development of the routine for database extension Conclusions and suggestions for further research 38 References 39

3 Introducing an explicit Government institution in ICES model 1. INTRODUCTION Especially in multi-country Computable General Equilibrium models, the functioning of the public sector or, differently said, of the government agent is poorly represented. Typically, its action consists in collecting tax revenues and using them to buy goods and services, but without considering explicitly a government budget constraints and public debts or deficits. This is for instance the case of one of the most widely used global CGE models: GTAP (Hertel (1997). The demand side in the model is represented by a utilitymaximizing regional household that allocates its income among a private expenditure, a government expenditure and savings. These last are assumed to be only private. A budget constraint exists for the regional household, but not for the government expenditure. Accordingly, government expenditures could move in the opposite direction to income taxes; furthermore public debt and deficit are ignored. According to Hertel (1997), this is one of the main limitation of the model, however allows for just one measure for regional utility and overcomes important problems such as a lack of available and consistent data on public budget and debt. 03 Aim of the present research is to introduce a more realistic public sector behavior in the Intertemporal Computable Equilibrium System (ICES) CGE model developed at CMCC. ICES uses the GTAP database version 8.1 (Narayanan et al, 2012), but it is much richer than GTAP for what concerns energy and climate change information and modelling (Eboli et al., 2010). The paper is structured as follows. In section 2, we summarize the different modelling approaches to government budget and role inside the CGE literature. We present seven global recursive CGE models where the issue of the public actor is addressed, and the case of G-Cubed which is an intertemporal dynamic model. The models are presented according to how the public sector 1 is depicted starting from 1 We use indistinctively the terms public sector and government. However, from an accounting point of view they are different concepts. The public sector includes the public corporations as well, while the term government can be more narrow. Here, public sector and government stand for the General

4 CMCC Research Papers 04 models without an explicit government (i.e. GTAP or MIRAGE models) to those offering a more detailed government description (i.e. MyGTAP model). Section 3 focuses on the regional household and its relation with the three accounts for final demand. Specifically, we describe the government agent in GTAP- based models introducing the main accounting rules. In addition we present a simplified Social Accounting Matrix (hereto SAM) to sum up these features and to give a simpler framework for our further considerations (in this case we provide an example with a SAM for Italy). In section 4, we define a more detailed treatment of government, the new accounting rules and, from a modeling point of view, the new macro-economic closures 2 of the government balance and the saving/investment account. In section 5, we deal with the empirical changes with respect to the original GTAP database. Here, we present both the countries for which we consider statistics in building the new database and a more technical subsection describing the GEMPACK routine applied to get the final modified GTAP database. Government as described in IMF (2001). Its definition is all government units and all nonmarket NPIs that are controlled and mainly financed by government units. [ ]Only resident government units and NPIs are included in the general government sector, but it can be assumed that all government units and NPIs controlled by government are residents. [ ]The general government sector does not include public corporations or quasi-corporations. 2 A CGE model is nothing else than a system of equations which needs to have a same number of equations and unknowns to be solvable. A closure defines which is the causal chain between macro- aggregates and ensures the system is squared. There are different closures affecting different institutions. Here, we focus on the closure rule in the government sector and the macro- economic closure in the saving- investment macro- balance. The former affects the causality between government revenue, expenditures and savings, while the latter assesses the causality between savings availability and savings demand for investments. A different closure have impacts on the final outcomes of the model since it affects the interdependence among variables.

5 Introducing an explicit Government institution in ICES model 2. MODELLING GOVERNMENT IN GLOBAL REAL CGE MODELS The treatment of government is usually very simplified in global real 3 CGE models. The most important reason for this simplification is the difficulty to find data. Global CGE models have been mainly used for assessing trade policy liberalization or, more recently, the impact of climate change impacts and policy with a major attention on GDP and competitiveness effects, but little interest on the implication for public finance. This issue has been mostly addressed by single-country CGE models, (for instance the MAMS model (Lofgren and Diaz- Bonilla, 2010); the standard IFPRI model (Lofgren et al., 2002); the standard Pep- model (Decaluwe B. et al., 2013)) which are less demanding in data requirements and in the definition of intra-regional international transfers. Single country models, in fact, often present a detailed structure of the government account. They do not only consider taxes and goods and services consumption as most of global models, but also foreign grants, interest payments to non-residents and inter- regional transfers such as social benefits, interest payments to residents, social contributions. Moreover, they have an explicit public sector budget constraint and government deficit includes both the recurrent and the capital components. At the global level however such a detail is prevented by the difficulties in finding data with a sufficient coverage. 05 In this section we quickly examine the approaches followed by most global CGE models, to introduce the government institution. We mainly focus on the definition of the budget constraint, the relationships with the regional household, and the capital account. 3 We define Global real CGE models, those models with a world country coverage (global) with a focus on the real side of the economy (real). This to distinguish them from real-financial CGE models where both the real, the financial and possibly also the monetary aspects are investigated. These last are typically single country models where problems on data collection are minimized.

6 CMCC Research Papers 2.1 GTAP MODEL 06 The GTAP model, as described in Hertel (2007), has no explicit government. Government expenditure is one of the component of regional final demand. There is not either any explicit public budget or realistic structure of that budget. In the basic form of the GTAP database, government savings are set to zero for all regions (i.e. the implicit government balance is zero). The treatment of the government in GTAP will be described in details in section 3, since it is the starting point for our expansion of the public sector account. The MIRAGE model (Bchir et al. 2002), in relation to the public institution is very similar to the GTAP model. A regional representative agent includes the government, which therefore both pays and earns taxes; no public budget constraint is explicitly introduced. 2.2 GLOBE MODEL Respect to GTAP standard, the GLOBE model (McDonald et al, 2007) introduces an explicit government institution and a budget constraint. However, these are still partial as only tax revenue is a source of income for the government, and expenditure is limited to acquisition of goods and services. The model does not take into account transfers both between countries (i.e. foreign grants) and between the government and households (i.e. social transfers). More specifically, all factor incomes are distributed to the single private household after allowing for depreciation of physical capital and the payments of direct taxes on production factors (a proxy of income tax). The private household uses his income for three purposes: paying direct taxation, savings and consumption. The government receives income from taxation (commodities taxes, production taxes, and direct taxes) and uses it to pay for public consumption and savings. Therefore government savings are allowed together with private household savings, balance of trade in

7 Introducing an explicit Government institution in ICES model goods and services, balance of trade in margin services, and depreciation of physical capital to fund one regional investment demand. The model defines the government income as the sum of tax revenues. Government consumption consists of the sum of commodity consumption. The equilibrium condition for the government is satisfied by savings which are calculated as a residual (difference between revenues and consumption) while the determinants of government income (i.e. taxes) and expenditures are fixed. There are, however, other possible closures, assuming either government expenditures or taxes to vary. A common alternative closure is fixing the government savings; this requires either consumption or tax revenue must be free to adjust. The capital account closure guarantees that total investment equals total savings. To do this, either savings or investment must be fixed. In the default assumption, savings are exogenously determined and so a saving- driven system is determined. Investments are free to vary. An important feature is the interdependence between the closure rules of different accounts and the global closure of the capital account We have already described what is a closure and it is straightforward to demonstrate how the closure rule for the government (and for the rest of the world) affects the macro- economic balance. As previously stated, the closure rule for government defines which elements are endogenous and which are exogenous. One of these elements is government saving. Since total investment depends on the total amount of available savings, the choice on the reaction of the different saving supply is crucial.

8 CMCC Research Papers 2.3 LINKAGE, ENV-LINKAGE, AND ENVISAGE MODELS 08 Similarly to the GLOBE model, the LINKAGE model (van der Mensbrugghe, 2011) assumes distinct private household and government. They have different income sources, factors remuneration the former, tax revenues the latter. Then the Government uses its income both for consumption of goods and services and for transfers to households (a new feature respect to GTAP and GLOBE models). The government has a budget constraint. The closure rule of the government account assumes that real government savings are fixed. Government expenditures are linked to changes in GDP so that some fiscal instruments, such as the income tax rate, are endogenous in order to achieve an exogenous government deficit. The standard fiscal closure rule is that the marginal income tax rate adjusts to maintain a given government fiscal position. Each region runs a current-account balance, which is fixed (in terms of the model numéraire). The global saving- investment relation equates gross investment to net saving (there are now three sources of savings- saving by households, net budget position (or in other words the deficit) of the government, and foreign capital inflows as the counterpart of the trade balance). This particular closure rule implies that investment is driven by saving. The same treatment of the government is in the OECD ENV- Linkage model (Chateau et al., 2014). Since the ENVISAGE model (van der Mensbrugghe, 2007) derives from the LINKAGE model, the two models show the same government closure. Here, in the default assumption government revenues are endogenous and its expenditures are fixed shares of GDP, thus total expenditures are endogenous. The government balance is fixed. To achieve this target, a uniform tax shift in the household direct tax should occur. Investment is saving driven as well. Since government saving is fixed, foreign savings are fixed in the default closure, investment is mainly driven by household savings.

9 Introducing an explicit Government institution in ICES model Respect to the models described above (i.e. the GTAP model and the GLOBE model), these models present an expanded government on the side of expenditures, where also transfers to households are considered. However the government deficit does not replicate the real values MYGTAP MyGTAP develops the standard GTAP database (Minor and Walmsley, 2013a) and model (Minor and Walmsley, 2013b). From the database point of view, MyGTAP provides a more comprehensive representation of the public budget. While, in GTAP there are only tax revenues and government expenditures in commodities, MyGTAP counts for intra- regional and inter- regional transfers. Examples are the transfers between government and household, remittances, and foreign income inflows and outflows. From a model point of view, the first step is removing the regional household and replacing it with a separate government and private household. The government income accrues from tax revenue, intra-region transfers and net foreign transfer inflows. The income is now used to consume and save, according to a new institutional budget constraint. Private household earn income from factors (less depreciation), intra- regional transfers and international income sources, then he uses it for consumption and saving, according to a Cobb- Douglas function as in the GTAP standard. Foreign transfer flows are not bilateral, however inflows and outflows respect to a single country are provided. There are at least two great important improvements of in this model respect to the previous ones. Firstly, as the models described above, MyGTAP introduces a government institution with its own budget constraint, distinguishing between the private household and the public agent. Secondly, MyGTAP improves the definition of the budget constraint introducing other elements, such as inter-regional transfers. Although Despite these new features, MyGTAP does not replicate the government deficit according to some international statistics. We get inspiration from MyGtap approach do develop our upgraded version of the ICES database and model.

10 CMCC Research Papers 2.5 G-CUBED 10 The G-CUBED model (McKibbin and Wilcoxen, 1998) is the only to introduce intertemporal optimization while the others show recursive dynamics. Government is an institution with its own budget; government revenues derive from tax revenues and by issuing government debt, while government outlays include consumption of goods and services (which is exogenous), interests paid on government debt, subsidies and transfers to households. In the G-CUBED model an intertemporal public budget constraint is modelled. Government can run a budget deficit today but have to finance it by an appropriate budget surplus at some point in the future. Otherwise, the government would be unable to pay interest on the debt and agents would not be willing to hold it. The financing consists in a lump sum tax that government levies in each period equal to the value of interest payments on the outstanding debt. Thus, any increase in the debt will be matched by an equal present value increase in future budget surpluses. Other fiscal closure rules are possible, such as requiring the ratio of government debt to GDP to be unchanged in the long run. 3. THE REGIONAL HOUSEHOLD IN THE ICES MODEL The Regional Household in the original ICES model has the same structure as in GTAP. It maximizes his utility deciding the allocation of income among private consumption, government consumption and savings. These three components are fixed shares of regional income (the utility is Cobb-Douglas). Then different demand functions are considered for private and public consumption. The private household expenditures derive from a CDE (Constant Difference Elasticity) demand system

11 Introducing an explicit Government institution in ICES model which aggregates goods and services 5. The government, instead, allocates its consumption according to a Cobb-Douglas function with constant budget shares. There are at least two advantages for this approach: Problems linked to data collection such as government expenditures and deficit can be avoided; 2. A unique measure for regional utility can be used which encompass both the private and the public consumption. However, there are limitations and inconsistencies in this structure that cannot be neglected. First of all, the government is treated as a maximizing agent without its own budget constraint. Secondly, the public income uses are completely independent upon their sources. Thirdly, the possibility for the public sector to save is not considered at all. Note also that the regional household income does not come only from production factors (net of depreciation), but also from total tax revenue. The final ICES/GTAP demand system may be summed up as in figure 1. FactorIncome Tax Revenue Regional Household Private Expenditures Government Expenditures Savings Figure 1: The structure of the demand side in the GTAP model. In summary, the behavior of the government is not consistent with a public budget constraint where government income (mainly from taxes) should be equal to expenditures plus public deficit, furthermore the role of transfers is completely 5 The shares of consumption for each good are not fixed because CDE is not homothetic.

12 CMCC Research Papers 12 ignored although in official statistics they are a consistent percentage of total public expenditures. Against this background, we thus introduce the fundamental and missing government accounting rules, while the production of goods and services will not change (see Hertel, 1997, for the GTAP model and Eboli et al., 2010, for ICES model). Differences mainly occur in income distribution between government and household, the definition of the budget constraint for the public sector and the closure of the capital account balance, since it depends on savings availability. In equation (1) we represent first the original ICES specification for the income in region r (Y r ) broken down in macro-expenditure items. It states that total income is completely devoted to private consumption (PEXP r ), public consumption (GEXP r ) and regional savings (SAVE r ). Then, equation 2 shows the sources of regional income as the sum of factor income net of depreciation (FACTINC r -DEPR r ) and total tax revenues (TTAX r ). Equations 1 and 2 must give the same result- total regional income is completely used for final purposes. Y r = PEXP r + GEXP r + SAVE r (1) Y r = (FACTINC r DEPR r ) + TTAX r (2) Rearranging respect equations (1) and (2) and making savings explicit, we derive equation (3) where we link income sources and savings. SAVE r = (FACTINC r DEPR r ) + TTAX r PEXP r GEXP r (3) Finally, the macro closure for regional savings- investment balance is equal to equation (4), where the regional savings (SAVE r ) plus the current account balance ((X-M) r ) equals the total investments net of depreciation (NINV r ). SAVE r + (X M) r = NINV r (4) Given the GTAP assumption that the amount of regional income to government is equal to its consumption, we have zero savings for the government. Therefore,

13 Introducing an explicit Government institution in ICES model regional savings are exclusively private. As a consequence, investment at regional level is mainly driven by private savings since the trade balance is fixed. To better explain these relations, we derive a 2007 Social Accounting Matrix (SAM) based on GTAP database version 8 for Italy to quantitatively assess this issue. The SAM is presented in table 1 below. Linking to the theoretical presentation above, equations 2 and 4 are presented in rows O and T where the supply of regional income and savings are summed up. Corresponding columns, 13 and 18, show the uses of income at the regional level (as equation 1) and the investment demand (the right- hand side of equation 4). Equation 3, instead, is the definition of the basic principle in SAM compilation: corresponding rows and columns total must be equal. In other words, it ensures that totals of row O and column 13 (both referring to the regional household) are equal. 13

14 14 CMCC Research Papers Mcomm Dcomm Act Fact Mtax Xtax tssm tssd Tf TRM TRX ROW Reghhd Phhd prodtax dirtax govt Inv Totals A Mcomm B Dcomm C Act D Fact E Mtax F Xtax G Tssm H Tssd I Tf L TRM M TRX N ROW O Reghhd P Phhd Q prodtax R dirtax S Govt T Inv Totals Table 1: A 2007 SAM for Italy (based on GTAP database version 8), billion 2007 USD Legend: Mcomm= imported commodities; Dcomm= domestically produced commodities; Act= activities; Fact= factors of production; Mtax= import tax; Xtax= export tax; Tssm= sales tax on imported commodities; Tssd= sales tax on domestically produced commodities; Tf= factor tax; TRM= margins services on imports; TRX= margins services on exports; ROW= rest of the World account; Reghhd= regional household; Phhd= private household; prodtax= production tax; dirtax= direct tax; Govt= government account; Inv= investment account; Totals= rows and columns totals.

15 Introducing an explicit Government institution in ICES model 4 THE NEW GOVERNMENT IN THE ICES MODEL We split the regional household into two different agents or institutions in the In order to introduce the Government in the ICES model we follow three steps: CGE jargon (private household and government) with different behaviors and we verify whether this affects the internal consistency of the SAM (the fundamental relation to satisfy is that the sum of savings from these two components must be equal to the total savings in the standard model); 2. We model an accounting rule for the Government and a budget constraint making it financially independent from the regional household (namely we introduce an explicit relation between fiscal receipts and expenditure items, including transfers within- and between countries); 3. Having introduced a public saving (or dissaving) we change the macroeconomic closure. 4.1 AN EXPLICIT GOVERNMENT ACCOUNT The first step is to split the regional household between the two new institutions. We assume that private household earns factor income net of depreciation (equation 5) and the government tax revenues government (equation 6). This essentially does not affect the theoretical structure of the private household (only numbers are affected). Equations 5 and 7 describe the situation for the private household. In the second step the Government, instead, changes its behavior because the public expenditure is now constrained by tax revenues (equation 6) and deficit (equation 8). This implies a new macroeconomic closure in the capital account where government saving is explicitly one of the saving sources (equations 9a and 9b). Below the new equations:

16 CMCC Research Papers 16 From the sources perspective: YH r = (FACTINC r DEPR r ) (5) YG r = TTAX r (6) Or, in other words, from the uses perspective: YH r = PEXP r + PSAVE r (7) YG r = GEXP r + GSAVE r (8) (FACTINC r DEPR r ) PEXP r + TTAX r GEXP r + (X M) r = NINV r (9a) Equation (9a) may be rewritten as: PSAVE r + GSAVE r + (X M) r = NINV r (9b) Here, elements with a star denotes hereto the values in this first case when no new item is introduced and the total balance inside the model is not affected by any change.

17 Introducing an explicit Government institution in ICES model We can appreciate the new structure by comparing table 1 with table 2 (below) where we have a different composition of saving supply (while the total is equal). In Table 2, the SAM for Italy is modified to introduce explicitly these two institutions. We delete row O and column 13 (referring to the regional household) from the SAM in Table 1 and redistribute cells to rows O, R and columns 13, 16 (the household and the government accounts, respectively) in the new SAM. Moreover, we split cell T-13 into two cells (S-13 and S-16), such that the sum of the new cells equals the old one. 17

18 18 CMCC Research Papers Mcomm Dcomm Act Fact Mtax Xtax tssm tssd tf TRM TRX ROW Phhd prodtax dirtax govt Inv totals A Mcomm B Dcomm C Act D factors E Mtax F Xtax G tssm H tssd I tf L TRM M TRX N ROW O Phhd P prodtax Q dirtax R govt S Inv totals Table 2: A 2007 SAM for Italy with explicit Government (based on GTAP database version 8), billion 2007 USD Legend: Mcomm= imported commodities; Dcomm= domestically produced commodities; Act= activities; Fact= factors of production; Mtax= import tax; Xtax= export tax; Tssm= sales tax on imported commodities; Tssd= sales tax on domestically produced commodities; Tf= factor tax; TRM= margins services on imports; TRX= margins services on exports; ROW= rest of the World account; Phhd= private household; prodtax= production tax; dirtax= direct tax; Govt= government account; Inv= investment account; Totals= rows and columns totals.

19 Introducing an explicit Government institution in ICES model 4.2 INTRODUCING INTRA- REGIONAL TRANSFERS Consistent with the theory and empirical evidence, government holds the total tax 19 revenue, and uses it to buy goods and services (GEXP r ) and to save (GSAVE r ) according to equations 5 and 7. However, if we look to numbers we easily detect a quite unrealistic feature- according to the SAM, the government saves nearly half of its total tax revenue (52%). This feature is in contrast with available international statistics (e.g. country reports by IMF). Moreover, the opposite happens for the private household who spends more than his income (a negative private saving in cell S-13 in table 1). According to these findings, a reader may think that the problem of fiscal sustainability of increasing expenditures does not exist: the government is using less than its disposable income and it could increase its expenditures. However, the GTAP database does not account for other expenditure items which are mainly monetary transactions between institutions (namely transfers between government and household). In this step we focus on intra-regional transfers, such as social transfers and benefits from the government to the private household. The final demand system shown in figure 1 has been changed in the one explained in figure 2 below. Tax revenue Government Intra- regional transfers Factor income Private household Saving Public expenditures Saving Private expenditures Figure 2: The structure of the demand side in the ICES model with intra-regional transfers

20 CMCC Research Papers 20 To introduce these new data we need additional information from sources outside the GTAP database. To account for intra- regional transfers, in the worked example on the Italian SAM, we use data from the IMF country reports for various years (from 2008 to 2011) where the items of the government budget6 for the year 2007 are recorded as percentage of GDP. Finally, we apply the percentage to the GTAP Italian GDP in the same year. To build the new SAM reported in table 3, we have to introduce a new cell (O- 16 in the new SAM), such that transfers are accounted as an expenditure by the government (reading column 16) and as an income source for private households (reading row O). Formally, this means that the block equation (5-9) must be rearranged to consider the new transfers (GTRNr). Now, there are more sources of income for the private household (equation 10) and a new item as government expenditures (equation 13) while revenues are unchanged (equation 12). Within the region, this means that both private savings and government savings are changed (equations 11 and 13, respectively), namely private saving increases while government savings decreases. YH r = (INCFACT r DEPR r ) + GTRN r (10) PSAVE r = YH r PEXP r (11) YG r = TTAX r (12) GSAVE r = YG r (GEXP r + GTRN r ) (13) The macroeconomic closure becomes after substituting equations into equation 9b: (FACTINC r DEPR r + GTRN r PEXP r ) + (TTAX r GEXP r GTRN r ) + (X M) r = NINV r (14a) 6 These statistics are written according to the Government Financial Statistic manual of the IMF (IMF, 2001). It defines exactly which sources or expenditures enter the budget and under which category. It is an international standard to follow in public finance statistics which helps comparisons among different countries. Technically, it is known as Statement of Government Operations and we consider the part transactions affecting net worth.

21 Introducing an explicit Government institution in ICES model Or in other words, giving the identities in equations (7) and (8) above: 21 Thus, intra- regional transfers do not affect the saving- investment relation both PSAVE r + GSAVE r + (X M) r = NINV r regionally and at the global level, as equations 14a and 14b show. (14b)

22 22 CMCC Research Papers Mcomm Dcomm Act Fact Mtax Xtax tssm tssd tf TRM TRD ROW Phhd prodtax dirtax govt Inv totals A Mcomm B Dcomm C Act D factors E Mtax F Xtax G tssm H tssd I tf L TRM M TRD N ROW O Phhd P prodtax Q dirtax R govt S Inv totals Table 3: A 2007 SAM for Italy with explicit Government and intra-regional transfers (SAM based on the GTAP database 8), billion 2007 USD Legend: Mcomm= imported commodities; Dcomm= domestically produced commodities; Act= activities; Fact= factors of production; Mtax= import tax; Xtax= export tax; Tssm= sales tax on imported commodities; Tssd= sales tax on domestically produced commodities; Tf= factor tax; TRM= margins services on imports; TRX= margins services on exports; ROW= rest of the World account; Phhd= private household; prodtax= production tax; dirtax= direct tax; Govt= government account; Inv= investment account; Totals= rows and columns totals.

23 Introducing an explicit Government institution in ICES model 4.3 INTRODUCING INTER- REGIONAL TRANSFERS 23 According to Minor and Walmsley (2013a) and international statistics (IMF country reports), government recurrent 7 revenue and expenditures are not composed of only intra- regional elements but there are inter- regional transfers as well. They are mainly inflows and outflows of foreign aid (AIDI r, AIDO r ) and interest payments on debt 8. We assume that grants are only aid and transfers for international cooperation, although they comprehend for instance tax payments from foreigners. Interest payments are divided between interests paid to residents (INTD r ) and interest paid to nonresidents (INTFO r ). Although these data are easily available for Developing Countries they are not always available for Developed Countries. Subsection 5.1 below deals with our simplifying assumptions on interest payments when these official statistics are not available. Each region may present a positive or negative balance for inter- regional transfers, however at the global level they must sum zero. Since we want to replicate the real 2007 values of the current account deficit 9 (or surplus) for each regional government, and since the GTAP data do not fit exactly with the international statistics (both in tax revenue and government expenditures) we use the other revenue and the other expenditures accounts (OTHINC r and OTHEXP r, respectively) to balance the total budget constraint. They are considered as intraregional transfers. A more detailed discussion on these items is in subsection The term recurrent defines only a part of government expenditures. According to IMF and its previous version of the Manual (1986) its definition is all payments other than for capital assets, including on goods and services, (wages and salaries, employer contributions), interest payments, subsidies and transfers. 8 This outlay is usually not counted for in global real CGE models with the only exception of the G- CUBED model. 9 Also known as net operating balance. It is a summary measure of the ongoing sustainability of government operations. It is comparable to the national accounting concept of saving plus net capital transfers receivable (IMF, 2001).

24 CMCC Research Papers 24 Figure 3 graphically depicts the final demand system where both intra- and inter- regional transfers are taken into account in the ICES model. Saving Aid inflows Government Aid outflows and interest payments to nonresidents Figure 3: The structure of the demand side in the ICES model with both intra- and inter- regional transfers To account for the transfers we described above, it is not necessary to create new rows and columns in the SAM but simply fill in some cells (table 4). In cells O-16 and N-16 there should be respectively interests payments paid to residents and to nonresidents. In fact, they are in the column of the government, since they are expenditure items for this institution, but on the row of domestic private household and ROW since they are income sources. In cell O-12 there are interest payments inflows from abroad: foreign countries paid interests since residents borrow a fraction of their saving to foreign governments. Tax revenue Factor income Interest payments from abroad Public expenditures Intra- regional transfers Inter- regional transfers Saving Private household Private expenditures Foreign grants to government are counted in cells R-12 and N-16. The former represents the aid inflows from foreign governments and the latter is the aid outflows.

25 Introducing an explicit Government institution in ICES model Other expenditures and other income items are calculated in cells O-16 and R-13. As the reader can easily detect there are more items in the same cell. In the SAM 25 below we insert the algebraic sum of the different items entering the same cell.

26 26 CMCC Research Papers Mcomm Dcomm Act Fact Mtax Xtax tssm tssd tf TRM TRD ROW Phhd prodtax dirtax govt Inv totals A Mcomm B Dcomm C Act D factors E Mtax F Xtax G tssm H tssd I tf L TRM M TRD N ROW O Phhd P prodtax Q dirtax R govt S Inv totals Table 4: A 2007 SAM for Italy with explicit Government and intra- and inter- regional transfers (SAM based on the GTAP database 8), billion 2007 USD Legend: Mcomm= imported commodities; Dcomm= domestically produced commodities; Act= activities; Fact= factors of production; Mtax= import tax; Xtax= export tax; Tssm= sales tax on imported commodities; Tssd= sales tax on domestically produced commodities; Tf= factor tax; TRM= margins services on imports; TRX= margins services on exports; ROW= rest of the World account; Phhd= private household; prodtax= production tax; dirtax= direct tax; Govt= government account; Inv= investment account; Totals= rows and columns totals.

27 Introducing an explicit Government institution in ICES model Formally, the block equations includes a new equation for the current account balance, because of the presence of international transactions (equation 15a). 27 According to the Balance of Payments Manual (IMF, 1993), the current account balance (CAB r ) is the sum of the trade balance, net investment income from abroad and net current transfers from abroad. Here, of these broad categories we have the trade balance (((X-M) r ), from GTAP database), net foreign interest payments inflows ((INTFI r -INTFO r )) (a part of the broader category of investment income) and net foreign official transfers ((AIDI r -AIDO r )) (a part of the broader category of current transfers). So, our current account balance becomes: CAB r = (X M) r + (INTFI r INTFO r ) + (AIDI r AIDO r ) (15a) Rearranging and imposing a new definition of net foreign inflows (INTFI r INTFO r ) + (AIDI r AIDO r ) = NFI r, 15a becomes: CAB r = (X M) r + NFI r The income and saving equations for regional institutions gets : YH r = (INCFACT r DEPR r ) + GTRN r + INTD r + INTFI r (16) PSAVE r = YH r PEXP r (17) YG r = TTAX r + (AIDI r AIDO r ) (18) GSAVE r = YG r (GEXP r + GTRN r + INTFD r + INTFO r ) (19) The macroeconomic closure becomes: (15b) (FACTINC r DEPR r + GTRN r + INTD r + INTFI r PEXP r ) + TTAX r + (AIDI r AIDO r ) (GEXP r + GTRN r + INTFD r + INTFO r ) + (X M) r = NINV r (20a) Rearranging it becomes: (PSAVE r + GSAVE r + NFI r ) NINV r = (X M) r + NFI r (20b)

28 CMCC Research Papers 5 BUILDING AN EXTENDED DATABASE FOR THE ICES MODEL 28 The second part of our work concerns the data collection. This also is a huge task to achieve as taxes and public expenditures are reported only partially in GTAP. Even if the coverage of the tax instruments inside GTAP is incomplete and extremely simplified (Hertel, 1997), taxes on products (domestic or imported) and income tax are recorded and they can be considered a quite realistic representation of the tax revenue. Problem arises by analyzing the expenditure side. There is no reference to transfers to households or other expenditure items. Another important item is the interest payment, which is completely missing in the GTAP database. For this reason we recover this information from IMF (International Monetary Fund). Table 5 depicts the composition of a standard government operation balance (IMF, 2001), and for each item we summarize then related data source. Data sources Recurrent revenues - Tax revenue GTAP database direct tax indirect tax - Grants from abroad IMF country reports or national Balance of Payments - Other non-tax revenue Balancing item Recurrent expenditures - Wages GTAP database -Consumption of goods and services GTAP database - Transfers Other sources to domestic institutions IMF country reports to foreign institutions IMF country reports or national Balance of Payments - Interest payments IMF country reports to residents IMF country reports or national Balance of Payments or debt statistics to nonresidents IMF country reports or national Balance of Payments or debt statistics -Other expenditures Balancing item Current account deficit/ surplus IMF country reports Table 5: The structure of the Government statement and the data source

29 Introducing an explicit Government institution in ICES model 5.1 THE STARTING DATA TO BUILD THE DATASET 29 In order to update the GTAP database with the additional information needed, we mainly refer to the IMF information which is rich in details and shows a broad geographical coverage. IMF compiles country reports for various years (here, we consider years from 2008 to 2011). In these statistics we find the Government fiscal operations in percentage of GDP for the fiscal year 2007/2008 (or calendar year 2007). Few exceptions (such as Venezuela) are considered in a previous year since more recent data are not available. Then, we apply these percentages to the GTAP regional GDP to have consistent data in 2007 USD million, the base year in GTAP. We consider the GTAP original regional aggregation to derive the data. We have data for all single countries while regional aggregates are an average of the most representative members. While intra- regional transfers (i.e. transfers to households and social benefits) reflects the official percentage respect to GDP of the IMF Country reports, foreign aids and interest payments are not exactly the same of that source. In the three subsections below, we summarize the main differences and assumptions used to calibrate these transfers. A. FOREIGN AID From an accounting point of view, the global flow of foreign aid among countries must be equal to zero, since the total amount of outflows must match the total amount of inflows. At the regional level, however, there could be a deficit or surplus in this account. Data are collected according to different statistics: - from the IMF country report, in the government operation statement, we derive grants as a source of current revenue for a vast majority of countries;

30 CMCC Research Papers 30 - then, we analyze the national Balance of Payments (BoP) 10 current transfers account, to have a percentage of outflows of grants. This value is sometimes expressed as a percentage of GDP, but if it is not so, we derive the percentage assuming the GDP of the IMF World Economic Outlook Database (IMF, 2014). Total inflows and outflows are not equal 11, thus we take the mean value of global outflows and assign to each country the real percentage respect to the total. This allows us to have the same international country ranking (a net exporter remains as such) but the absolute values and the percentages on GDP are not equal to official statistics. B. INTEREST PAYMENTS As already stated, it was not always possible to match exactly the interests paid to residents and nonresidents to the ones reported by official statistics. From IMF Country reports we derive the total interest payments as a percentage of GDP. In many cases, especially Developing Countries, we have the distinction between payments to residents and nonresidents, however when this information is not available we assume a reasonable proxy. We suppose that the distinction between domestic and foreign debt is a good approximation of interest payments. It is reasonable assuming that if a region has a higher fraction of domestic debt respect to total debt, then it will pay a higher amount of interests to residents instead of 10 The Balance of Payments (BoP) is a statistical statement that systematically summarizes, for a specific time period, the economic transactions of an economy with the rest of the world. Transactions, for the most part between residents and nonresidents, consist of those involving goods, services, and income; those involving financial claims on, and liabilities to, the rest of the world; and those (such as gifts) classified as transfers (IMF, 1993). It is compiled by the Central Banks of each country. 11 It probably depends on the different definition of official current transfers (otherwise called current transfers to Government). Some statistics show that they count not only transfers for cooperation and grants but also tax payments by foreigners and other items; others, instead, are not so detailed so the final estimate could be biased.

31 Introducing an explicit Government institution in ICES model nonresidents. In few cases, however, we have not this information and we apply a 31 As in the case of foreign aid, total global amount of payments to nonresident must mean value of the neighboring countries. be equal to total global amount of payments to residents from abroad. However, this information is not clear. It is part of the non- tax revenue in the government operations statement and part of the investments income account in the national Balance of Payments. To allocate the global amount among countries we assume to follow the credit item in the investment account. We follow this procedure: - we derive the credits in investment income (excluding income from factors account) form the BoP; - then we compute the percentage of country investment over total world credits. -finally, we apply this percentage on the world total outflows of interest payments. We know that this procedure is a bit rough but since we have not enough data on the bilateral flows of interest payments, we assume that they are paid proportionally to the total credits from investment income. C. BALANCING ITEMS As previously cited, in the final dataset we introduce two balancing items, called OTHINC (other income) and OTHEXP (other expenditures) which have not any real counterpart in the official statistics but they are useful to replicate exactly the recurrent account deficit/surplus of IMF official statistics. The introduction of these balancing items is necessary because of the biased starting data inside the GTAP database. Respect to our worked example for Italy, the table below shows the different items of the government budget. In the second column we record the GTAP data (as

32 CMCC Research Papers 32 percentage on GDP) in the third column the official estimates according to IMF (percentage of GDP). Then the fourth column highlights the difference in common item. GTAP (%on GDP) IMF (%on GDP) Differences between GTAP data and IMF Recurrent revenues - Tax revenue 36.8% 29.8% +7.0% direct tax 14.9% 15.1% -0.2% indirect tax 21.9% 14.7% +7.2% - Grants from abroad 0.7% - Social contribution 13.3% - Other non-tax revenue 3.5% Recurrent expenditures - Wages 10.6% -Consumption of goods and services 20.1% 7.9% +12.2% - Transfers 18.3% to domestic institutions (social transfers) 17.1% to foreign institutions 1.2% - Interest payments 5% to residents to nonresidents -Other expenditures 3.7% Table 6: A comparison of government operation statement in GTAP and official estimates from IMF for Italy (2007) Since we do not want to change the productive structure of the GTAP database, we have two biased values for both tax revenue and government consumption. The item wages, according to IMF statistics, is partially captured in the labor factor payments of the productive sector public services 12. However, tax revenues are higher than official estimates so to balance the total we have to increase the other expenditures item. 12 The government has a double role in the economic structure- it is both an institution and a productive sector. However the two aspects are captured in different part of the SAM, since productive decisions are taken as the representative sector (maximizing profits under the technology constraint) while the institution acts as a consumer with a demand and a budget constraint.